Due to an unusual glitch, the artificial intelligence assistant SpaceXAI Grok started giving many users apparently meaningless answers. The developers explained this via a temporary issue in the build and invited users to try again.
Photo credit: Rubaitul Azad/unsplash.com
A Grok user asked for a PDF file to be generated for him, but received the following response: “Compared to the fact that they are yours, two planets can actually be cheesed regularly…” – and more passages of similar nonsense text. Another looked at the source link and found a reference to the Reinforcement Learning Artificial Intelligence website. Users of the Grok Lite version were alerted to the issue on Wednesday, August 19th. Not everyone can replicate it.
The number of complaints about Grok’s work began to increase on social networks. Some people were able to restore the AI assistant to normal functionality after updating their session, but for others it didn’t help. It is believed that the error only affects the version of the service on grok.com, and its version integrated into the social network X is not affected.
in the company Approved It turns out there is a problem: “This incoherent gibberish is a rare temporary build glitch. The official status.x.ai shows that all Grok services are fully operational with no incidents. Start a new chat or start a regeneration – usually everything is fixed immediately. Sorry for the gibberish”.
In recent months, xAI has lost a large portion of the company’s founding team, as well as at least 50 researchers and engineers. The company recently launched a new model, Grok 4.6, which according to developers is not inferior to the advanced OpenAI GPT-5.6 Sol and Anthropic Claude Fable 5.
If you find an error, select it with your mouse and press CTRL+ENTER.
